roadmap
- 01
- 02
- 03
- 04
- 05
- 06
- 07
- 08
Prepare for software supply chain security interview questions on trusting code you did not write. Covers dependency risk, SBOM and VEX, signing with Sigstore, SLSA provenance, and pipeline hardening.
328 of 328 questions in your plan: 0 known, 0 in progress, 0 due, 0 skipped, 328 new. 0 cut, outside your plan.
The roles that ask about this tree each enter it through a different door. DevSecOps interviews live in the pipeline: what a workflow triggered by `pull_request_target` can reach once it checks out a contributor's branch, why a third-party action pinned to a tag is not really pinned, how a job proves its identity to a cloud account or a signing service without a stored secret. DevOps and SRE loops ask from the operator's side, closer to the registry: pulling a base image by digest rather than by tag, and refusing to deploy an image whose signature does not name the expected builder. Application and product security loops start from the dependency graph and end at policy. Expect a scanner report with a critical finding in a library function the service never calls, and the question of what you do with it; then the paper trail: what a VEX statement asserts, how an SBOM names a component without ambiguity, what the NIST SSDF or the EU Cyber Resilience Act expects of a software producer. AI red-teaming rounds reach the subject through what a model pipeline pulls in: a model file that runs code the moment it is deserialized, or a package name an assistant invented and an attacker then registered. For the whole interview rather than this subject, the [DevSecOps Engineer](/roadmaps/devsecops), [DevOps / SRE Engineer](/roadmaps/devops), [Cyber Security Expert](/roadmaps/cyber-security) and [AI Red Teaming](/roadmaps/ai-red-teaming) roadmaps set it beside the cloud, detection and attack questions those loops also ask.
Skim the [supply chain hub's primer](/topics/sec-supply-chain) before your first drill: it is the map, and with it each question lands somewhere you already recognise. After that, treat the plan as a bank of questions to drill, not a syllabus to read in sequence. When one of your answers names a control but not the attack it stops, that is the cue to reopen the primer. Juniors gain most from [dependency risk](/topics/sec-supply-chain-dependency-risk): what a lockfile guarantees and what it does not, how a vulnerability arrives through a transitive dependency, and why a finding in code your service never reaches can wait behind one it does. Middle candidates are usually separated in [pipeline hardening](/topics/sec-supply-chain-pipeline-hardening), where the attacker is a pull request or a compromised action and the answer turns on which token the job holds and what that token can write. Senior conversations settle on [provenance](/topics/sec-supply-chain-slsa-provenance): how a signature, which ties an artifact to whoever released it, differs from an attestation of how it was built, and what a verifier has to check before either means anything. A keyless signature verified without pinning the expected identity and issuer proves only that somebody signed. Seniors with a loop days away can lift the floor of level off junior: what a lockfile is, or what SCA stands for, is assumed at their band, and the hours that frees belong to the pipeline and provenance questions.
The roles that ask about this tree each enter it through a different door. DevSecOps interviews live in the pipeline: what a workflow triggered by pull_request_target can reach once it checks out a contributor's branch, why a third-party action pinned to a tag is not really pinned, how a job proves its identity to a cloud account or a signing service without a stored secret. DevOps and SRE loops ask from the operator's side, closer to the registry: pulling a base image by digest rather than by tag, and refusing to deploy an image whose signature does not name the expected builder.
Application and product security loops start from the dependency graph and end at policy. Expect a scanner report with a critical finding in a library function the service never calls, and the question of what you do with it; then the paper trail: what a VEX statement asserts, how an SBOM names a component without ambiguity, what the NIST SSDF or the EU Cyber Resilience Act expects of a software producer. AI red-teaming rounds reach the subject through what a model pipeline pulls in: a model file that runs code the moment it is deserialized, or a package name an assistant invented and an attacker then registered.
For the whole interview rather than this subject, the DevSecOps Engineer, DevOps / SRE Engineer, Cyber Security Expert and AI Red Teaming roadmaps set it beside the cloud, detection and attack questions those loops also ask.
Skim the supply chain hub's primer before your first drill: it is the map, and with it each question lands somewhere you already recognise. After that, treat the plan as a bank of questions to drill, not a syllabus to read in sequence. When one of your answers names a control but not the attack it stops, that is the cue to reopen the primer.
Juniors gain most from dependency risk: what a lockfile guarantees and what it does not, how a vulnerability arrives through a transitive dependency, and why a finding in code your service never reaches can wait behind one it does. Middle candidates are usually separated in pipeline hardening, where the attacker is a pull request or a compromised action and the answer turns on which token the job holds and what that token can write. Senior conversations settle on provenance: how a signature, which ties an artifact to whoever released it, differs from an attestation of how it was built, and what a verifier has to check before either means anything. A keyless signature verified without pinning the expected identity and issuer proves only that somebody signed.
Seniors with a loop days away can lift the floor of level off junior: what a lockfile is, or what SCA stands for, is assumed at their band, and the hours that frees belong to the pipeline and provenance questions.